Computer Networks

Prof. Sujoy Ghosh

Department of Computer Science & Engineering
Indian Institute of Technology, Kharagpur
Lecture - 1
Emergence of Networks & Reference Models
Good day. I am Professor Sujoy Ghosh of IIT Kharagpur; I teach in the Computer Science and
Engineering Department, I will be taking this course on computer networks. We will be having
about 40 lectures in the series. Today, we will start with an introductory lecture, which is about
the Emergence of Computer Networks, its brief history, and a little bit about the protocols and
reference models, which is an abstract view of computer networks. The text books for the course
will be these three: Computer Networking: A Top down Approach Featuring the Internet by
Kurose. Data and Communication by William Stallings and Computer Networks by an S
Tanenbaum. Some of the material that we will be covering here verbally as well as in the slides
will be from these books. The first question is what is a computer network? Computer network is
a number of computers (also known as nodes) connected by some communication lines.
(Refer Slide Time: 00:36)

We have these computers, which are connected by some communication lines. Two computers
connected to the network can communicate with each other through other nodes if they are not
directly connected, which means that computer does not have to be connected to all the other
computers in the network in order to communicate with any of them. It may be an indirect
communication via some other computer and other point is that these nodes are computers. But
some of them are not computers; some of them are network devices. There are various kinds of
network devices like switches, routers etc., so, they facilitate the communication and running of
this entire network.

(Refer Slide Time: 1:28)

They are also taken as nodes in the network. Finally, we have some of the nodes, which may be
either computers or they may be network devices like switches and routers and there is a
communication between them. This communication can be of various types we will come to
that presently. Can be of various types we will come to that presently. The computers can
communicate with each other. What do they communicate? They exchange information between
different computers, this information could be of any kind of information they could be data
used by a program or they could be some program itself or it could be any kind of information.
This is one of the basic and major uses of computer networks. There are other specialized uses of
computer networks. For example, interconnected, small computers can replace a large computer.
For example, you have a very large computation to perform and so for that either you need a
very powerful and very large computer, which is very costly. The other option could be that you
break up this work into very small pieces and assign them to the small computers. The small
computers do the computation on that small chunk of the problem, and then they communicate
with each other to form the final solution. The other use of computer networks, which is coming
into vogue very much these days, is that you can use this network as a communication tool. For
example, you can send emails to almost anybody these days, and that is really a very cheap and
very fast mode of communication.
Secondly, the computer networks could also be used for direct communication like you could
communicate through voice over a computer network; you can communicate through video over
a computer network. All these different communication is converging into this computer network
and nowadays when we talk about the network we do not look upon it just as a computer
network. It is the bedrock on which, all these computing communication are converging and then
you have some applications, which are necessarily of distributed nature. For example, railway
reservation system now obviously if you had a large computer to handle all the railway
reservations in the world in one place but we would not want all the people to go to one place
and form a huge queue. So, we want to distribute this functionality that means what that of
booking tickets all over the country.

These are the examples of distributed systems so this is a distributed reservation system.
Similarly, there could be other kinds of distributed systems, distributed databases and all kinds of
applications, so that is the other application of networks.
(Refer Slide Time: 3:30)

Now, a quick look at the history of the computers first, then we will focus on computer networks.
In 1948, we have the first commercial computer installed; it was a UNIVAC that was a big
company once upon a time, and then in 1958, the first US communication satellite became
operational. The network always has these two components, one is that you have to have these
computers and then you have to have these communication channels and satellite communication
is one of the important medium of communication. In 1964, we have an airline reservation
system, which uses some kind of a packet switching network; it was proposed by the RAND
Corporation; and in 1969, ARPANET, the first packet switching network began its operation.
This was actually a water shed event as we will see. These have been very far reaching impact on
the way computer networks developed.

(Refer Slide Time: 6:51)

In 1971, we have the first computer chip, previous computers were made through very low level
gates and all but the first four-bit small computer chip with 2300 transistors and they became
available in 1971. In 1972, Ethernet specification was formulated this Ethernet is one protocol
and Ethernet remains one of the most important network protocols that we use. So that was
first formulated in 1972. In 1974, IBM introduced its own version of network in SNA. In 1975,
Altair 8800, the first commercial microcomputer, was sold as a kit. In 1975, we had Paul Alan
and Bill Gates getting together who eventually formed Microsoft. In 1976, Woznaik and Jobs
built Apple1.
(Refer Slide Time: 8:17)

So Apple also has lot of contribution to make to computers in various aspects and Windows
system was basically started from Apple computers. In 1979, we had VisiCalc first commercial
spreadsheet; this is the popular application of computers these days.
(Refer Slide Time: 09:18)

In 1981, IBM introduced IBM PC with one floppy drive; later on they changed it with drive and
hard drive; and in 1983, TCP by IP is another network protocol that we are using. We use both
Ethernet and TCP by IP protocol. We have different protocols, which we are running on the
network and how this different protocols are arranged and how they are used that I will cover
today in second part of the lecture. So, this TCP by IP is another important protocol that is still
very much prevalent today that became the official protocol of the ARPANET. ARPANET is
the network, which was set up in USA under the DRDO and which had a very far-reaching
impact on the way the network developed. In 1984, Apple introduced GUI with Apple Macintosh
and 1986 we see the first laptop PCs.

(Refer Slide Time: 10:09)

In 1988, IBM gives a multitasking OS for the PCs. In 1989, Microsoft releases Windows, which
is yet another multitasking system that is the Microsoft version; in 1989, NSF, National Sound
Foundation of USA replaces ARPANET as internet backbone and then www was invented by
CERN Physicist Tim Burners Lee in 1991. So this advent of www, World Wide Web, made the
network very popular.
There was this kind of positive feedback going on, as it became more popular, more and more
people jumped into it, as people showed their interest, so commercial companies also became
interested because there would be a market for something in the networks that they do and when
the commercial companies developed something, naturally it became more easier to use the
network; they had more functionalities; So more people became interested, that was the positive
loop. Actually the growth of network in the past decade in the 90s has been absolutely
phenomenal. The basic so called killerapp, application which attracted so many people to
networks was this www which was invented in 1991.

(Refer Slide Time: 11:08)

In 1992, Mosaic released first GUI web browser, so that was the web browser then Netscape
became the popular web browser, Microsoft had its own version of Microsoft Explorer, that was
an another browser and then it all built up into some kind of a frenzy and in 2000, we had this
.com, so-called .com meltdown. That means people became absolutely frenzied about the growth
of computers and such frenzy cannot go on forever so it has to come to a halt. So, it did in 2000
but the development of networks and its utility and importance remains.
(Refer Slide Time: 12:48)

Now, let us go to a brief history, more focused on the network itself. We were in general talking
about the computer scene. In mid-1960s, USAs Department of Defense wanted a command and
control network that could survive a nuclear attack. If something had to survive a nuclear attack
then you had to distribute it. So that was the one strong reason that we wanted a distributed
system and if you have a distributed system, these systems must be able to communicate. So that
was the seed of this project. The subnet should consists of some Honeywell 12 KW intermediate
message processors so there are some kind of early network devices, which are connected by 64
Kbps lines. These IMPs Intermediate Message Processors are connected to the various
computers and the IMPs themselves were connected by telephone lines from telephone
companies.
(Refer Slide Time: 13:34)

Later, the IMP software was changed to Terminal Interface Processor, which allowed the
connection of terminals. In the 1970s, NSF setup a machine connected to the ARPANET to
which other universities could dial up and connect. Now, this became very important in the sense
that many people were interested in this so they upgraded their systems in various ways. In mid1980s NSF built a new backbone to connect with supercomputing centers to some regional
networks. Its backbone was upgraded to 448 Kbps and then to 1.5 Mbps fiber backbone, and
once this fiber backbone was in place, communication became really fast.

(Refer Slide Time: 14:37)

So it allowed network connection to thousands of universities, research labs, libraries, museums,

etc. As the growth continued, commercial houses began to notice and to join; the decentralized
model began to take hold as I described earlier, commercial people joined so there was a lot of
innovation by a lot of people.
(Refer Slide Time: 15:18)

One of the major driving forces in the computing networks in the past decades or even now may
be has been the role of so-called Start Up Companies. Start-up companies means that some
young engineers or some young people who had some very bright idea which they developed to
a certain extent and when that was proven and then some big company would possibly buy their
company or give them money to get access to this kind of technology. There were so many
success stories that a lot of people got interested in it and when a lot of people think about a
problem, there is a lot of innovation. So that is how computer networks had an explosive growth
in the whole of 90s and even in the 21st century and the growth is going on. Although the frenzy
in the stock exchanges, etc. about these .com companies crashed in a 2000 but the growth in the
field of networks and its use in various walks of life, is continuing. With this brief historical
background, now let us come to the networks and some kind of structure or some kind of an
abstract view of this computer network.
I have already mentioned that computer networks will be a number of nodes and nodes could be
computers or network devices, which are connected by some communication lines. Sometimes
things are not so simple; sometimes you have a line between two nodes, giving us point-to-point
communication. That means one node A is communicating directly to one another node B, where
A and B are connected by a direct line, it could be physically a copper line or fiber line, etc. This
point-to-point connection could be the simplest case but actually it is dedicated and in the
dedicated case also we have some different cases. For example, you could have simplex
communication; it means that communication can go in only one direction in that line.
It could be half duplex, it means the communication could go both ways either way from A to B
or from B to A, but only in one direction at a time, and full duplex, that A and B can
communicate with each other simultaneously, in parallel, at the same time. So that means there is
communication from A to B and B to A at the same time. .These are some different kinds of
dedicated lines. Then if the nodes are not computers, some network devices, specifically, if they
are multiplexers, we could share a point to point medium.
That means there is one line from A to B but a is connected to a lot of other nodes say, A1, A2, to
An and B is connected again to a lot of other nodes say, B1, B2 to Bn and they could all these Ais
and Bis could communicate through this one single line, which is between A and B, this we will
see later how such a thing can be done.
This business of sharing a link over time or whatever even parallels sometimes it is possible.
Sharing this link by multiple people is called multiplexing. These are the different kinds of pointto-point links but point-to-point links are not the only things we have. For example we could
have a broadcast kind of link if we have a satellite communication, satellite throws its signal all
over the country or may be all over the region. That is something, which is being broadcast so
you cannot really put it down as a point-to-point link between something, but you could use that
shared medium and a broadcast in some way to make temporary point-to-point links or your
application may be such that you want to broadcast something. For example, we broadcast TV
signals; similarly there are things, which we want to broadcast all over.

This broadcast media can be dedicated to some users or it could be shared; that means it is shared
between multiple users. There the term used is multiple access medium, to which different users
are connected at different points and it is a broadcast kind of medium. In a distributed fashion,
they have some protocol of deciding how to share this because if it is in the same place, the
business of sharing becomes somewhat easier so you can multiplex it. If nodes themselves are
physically distributed over the broadcast medium you will have some kind of protocol called
Multiple Access. So you can have a shared broadcast medium also; that is quite common.
Finally, we have something which is in between broadcasting and point-to-point, which is called
multicasting, this means that if we have a group, I have, say, some friends and I want to send
some things not to one particular friend but to my group of friends and to no others so
multicasting is basically communicating to a specified group.
(Refer Slide Time: 15:46)

Using such communication link, we have this computer networks. Networks are divided into
Local Area Network, Metropolitan Area Network, and Wide Area Network. Local Area Network
means that it is local, limited to maybe one building or small group of buildings so its size is
small. There are some other things about Local Area Network, which make the issues of Local
Area Network somewhat different from a Larger Area Network and the Local Area Network
usually could be privately owned. Now how does ownership come into picture?
Ownership comes into picture because if you have the same owner for the entire network then
you can have the same policy, which is not the case in wide area network. In Wide Area Network
there may be various nodes which are connected to this Wide Area Network. The nodes may
belong to very different people and they may have very different ideas about what should be
done and all their policies. That makes Wide Area Networking somewhat different from Local
Area Networking. Metropolitan Area Network means a network which is spread over a
community or maybe even a city, their size is bigger than Local Area Networks and one of the
issues which is very important is the access issues, which means that how do you connect?

For example, you want to make the entire community networked; so how do you connect each
one of them because they are geographically more distributed then a Local Area Network. Local
Area Network may be sometimes we would use validers or sometimes we would just simply
draw valve and the cost will not be very prohibitive but in local area how we connect the people
becomes an issue that is the access issues. We have talked about the Wide Area Network. Wide
Area Networks are costly they are communicating over may be hundreds or thousands of
kilometers. The communication is costly, the cost has come down and that is another thing,
which has historically happened.
As more and more people got interested in network, technology developed that is one side of the
issue. The other side of the issue is that volumes went up; which means that the number of
people who used network so as the volumes went up the cost went down. So the technology
improved in one direction. That was one input into bringing down the cost and the other issue
was that more and more people started using it. So the volume to a particular company
developing something in the network area went up and they could make things cheaper. And as
the whole thing became cheaper and cheaper, more and more people wanted to join the network.
Wide Area Network cost is still an issue. Wide Area Network may connect various Local Area
Networks, so its a network of networks or internetwork or internet.
(Refer Slide Time: 22:26)

Let us look at the abstract idea about the networks in general. In formal framework, we have
what are known as protocols, why do we need protocols in network? We need protocols in
networks because networks by their very nature might connect various different people as I said
various different people may have very different ideas about how things have to be done. If
people have different ideas about how things have to be done then they have to agree on some
common basis to communicate and this common basis is the protocol. There are numbers of
protocols, which are used in network and we will be getting into that.

You and I may have very different ideas about how some particular thing may be done. I may be
doing this in a certain way and you may be doing the same thing maybe in a completely different
way. But if we want to communicate we have to agree on some minimum protocol. These
protocols are the building blocks of this network architecture. So each protocol object has two
different interfaces one is the service interface that defines the operations on this protocol and
peer to peer interface which defines messages exchanged with peer. For example, any protocol, a
protocol is between peers so if in this diagram there are the two Lis are the peers and they
communicate with each other through this peer interface; that is the protocol.
This protocol is supposed to achieve something, supposed to do something. For this it requires
something maybe from some other called layer. Actually we have a layered architecture; I am
coming to that in the next slide. How this protocol is to be invade etc for this we have a service
interface and between the peers we have a peer interface.
(Refer Slide Time: 26:18)

We have all these different layers; we have this N plus 1 th layer, N th layer and N minus 1 th
layer. How they are to be layered about that also there are some standards and we just look at
two of them at least. For example, this networking business is broken up into whatever the jobs
you have to do for achieving for smooth functioning of computer network we break them up into
different functionalities and these functionalities are in different layers. This is all in one place
that means in one node. Similarly, another node, which is connected to the network, will have its
own layers and at the same level N and N here they are peers. So there will be a protocol
between this N and N, there will be a service interface between N plus 1 th layer and N th layer.
Similarly, there may be a service interface between N th layer and N minus 1 th layer and so on.

(Refer Slide Time: 28:51)

So let us see what they are like. As I said most networks are organized as series of layers. The
task of each layer is to give some service to the upper layer and any layer maintains a virtual
connection with the corresponding layer in a peer. Any layer maintains a virtual connection; this
virtual connection is used for running the peer-to-peer protocol, whereas this is the service
interface.
The task of each layer each layer performs certain subtask of this whole networking business,
so each layer performs some task is a service to the upper layer. Similarly for performing this
task, it may break it up into some subtask and some subtask may be delegated to a lower layer.
This layer could have its own service interface to the lower layer whereas its own task, assuming
that the subtask is done by the lower layer, is performed at the peer-to-peer level, using the
corresponding protocol.

(Refer Slide Time: 30: 03)

There is peer-to-peer protocol running between any two corresponding and communicating
layers. The interface between the layers in the same node is well defined. Another point is that
for this peer-to-peer protocol mostly they go through a virtual connection. What is a virtual
connection? Virtual connection means that physical connection; we understand let us say two
computers are connected by wire. It is not necessary that two computers have to be directly
connected in order to communicate; it may be an indirect connection also. That means it is going
from one computer to another then, it is hopping from node to node then, it is finally reaching its
destination, and so there is no direct physical communication.
When you are surfing the internet, you have been connected to some computer, which may be
connected to opposite part of the globe so, to you it appears that you are directly connected to
that computer so whenever you are clicking something over here, something happens. So that
clicking of the mouse somehow that gets communicated to that remote web server and may be
the page changes or something. There seems to be a direct communication but this
communication is only virtual.
The physical communication is only at the lowest layer, where physically something is
connected by a piece of wires or fiber optic links or by satellite links whatever it is, that is the
physical connection. For running all these protocols, these peers need to communicate and this
communication is through virtual connection. This is the main point of having this layer
architecture: that the implementation of each layer, each node is transparent to the other two
nodes. As I said, I may want to do this one particular subtask in some particular way I means
some company X and then some company Y wants to do the same thing in a different way.
Now you need not stop any of them so long as they agree on the protocol, so long as they agree
on the service layers. What is the service interface between the different layers because upper
layer service may be given by some product of some company, whereas the lower layer may be
developed by some other company. Once again if they can decide on the interface between these
two layers, then they are very free to do it in their own way. The same thing applies to the peers.

For example say two particular layers in two particular nodes may be peers, these nodes are
computers themselves. Now, one may be an IBM machine another may be a sun machine and
they will have different operating system, they will have different processors etc. But so far as
the peer to peer protocol is concerned, they agree. So these are the messages I will send and then
if I am expecting these kinds of answer, these kinds of messages, I will accept and these kind of
answer will be given etc, so that is how a protocol goes. So long as they agree on this protocol,
these two companies are free to develop their product in their own way and when you allow that
naturally, people can innovate; people can put in different things. So long as you are conforming
to the standards of the protocols and the interfaces, etc., you can develop your own thing in your
own way and that is really good for the development of the entire network.
The other point is that this is one way of abstracting out all these unnecessary details. For
example any of this network operating system in a computer or a networking device, they may
be very complex indeed. Some of the complexity is special to the implementer. One implementer
has decided to implement something in a particular way and that will have a whole lot of details.
We are not really interested in that one, we have to abstract out of that. We concentrate on these
protocols and the interfaces and the functionalities; that gives us a fairly general picture about the
entire networking, how networking is done and then if you go into the business of developing
some of these modules yourself, then you have to go into some more nuts and bolts about how
this service can be given.
The protocols between the peer layers can be changed if the peers all agree. Naturally if you are
changing the protocol, all the peers have to agree; otherwise the communication will breakdown.
However, it need not be referred to other layers so that is another good thing about layering;
that is whatever change I make it will not matter and the service definition says what the layer
does and nothing else, means nothing about the specifics of implementation, which may vary
from one vendor to another. The interface tells the process about how to access it; it specifies
what the parameters are and what results can be expected.

(Refer Slide Time: 31:16)

We have mentioned the three layered models; the most famous is the OSI reference model which
we will look into in some detail and then TCP by IP reference model which is most widely used.
OSI reference model mean some part of it is very somewhat theoretical because many people
really do not do that and it is that they do not consider that part to be very important. TCP by IP
reference model is something which is almost ubiquitous. And of course there is ATM, it is one
networking technology which is rather complex technology; as I mentioned here just as an
example; so that is another kind of reference model. There are all kinds of different reference
models, sometimes these models are basically a description of the different layers that are there;
and naturally if you talk about the layers, what the layer is supposed to do and that is what you
see and then how it is to be interfaced with the layer above and layer below. Since this is number
of layers, one on top of other, they are also referred to as stacks.
We call about this as TCP by IP stacks, OSI stacks, ATM stacks and so on. Other protocol stacks
exits and new ones are possible. However the extent to which a particular model is universally
accepted is the key to its success. As I said a lot of thought went into OSI reference model but in
practice TCP/IP became much more prevalent.

(Refer Slide Time: 37:58)

OSI reference model has seven layers and these layers are; application layer, the presentation
layer, the session layer, the network layer, the data link layer and the physical layer, which
means that in order that people can communicate over this network in a very seamless manner,
all the jobs that are involved have been broken down into seven parts. So they are different layers
application layer is something and then presentation, session, network, data link and physical
layer. Each layer has some kind of functionality and all these functionalities together give you
the overall functionality in the network.
(Refer Slide Time: 39:37)

Little bit more about peer-level communication: messages sent from one application to another
application on different host, travels down to the layers of the sending machine and each layer
adds a header to be used by its corresponding peer level and the bottom layer, which is the
physical layer, sends the message to the TCP machine; this is the general scheme. The point is
that suppose we start something at the very top, in the case of OSI, we call it the application
layer. Some communication has been initiated at the application layer. This application layer,
mind you, has its peer in the destination machine. In the destination machines also some
application layer program is running.
For example let us say you are doing a TELNET kind of thing, which means you are logging on
to another machine, TELNET to another machine. So you will start a TELNET program, which
is your TELNET client program and the destination machine will respond to a TELNET client.
Who will respond to TELNET client; only a TELNET server can respond to a TELNET client.
TELNET server is under the remote machine. These are the two modules which are in the
application layer. How TELNET is an example of a protocol? This is an application layer
protocol and what this protocol does, how the TELNET client will request and how the TELNET
server will respond etc. are the internals of TELNET protocol.
So far as the TELNET client or any of these application programs is concerned when it tries to
communicate with another application program in another machine it just knows about this
protocol and how things have to be done at that layer but how this communication is able to go
from this machine to another machine? For this, if you put the whole thing in the same program
that becomes very complicated; that was the idea of layering. So give him a virtual connection,
he will communicate to the target machine, he will have some data to send that will add up
maybe to other kinds of data and then hand it over to the lower layer in the same node.
The lower layer may be running a protocol with the corresponding layer in the remote machine.
For that protocol it needs to do communication, for that communication whatever message has to
be sent here that gets added to the original data, which was sent by the application layer. As the
message to be sent moves down the stack from the originating machine at each layer, each layer
is running its own protocol with its peer and it has some message to add to give to his pair. So,
they will keep on adding it to this and this becomes fatter and fatter as it goes down. As it goes
down to the physical layer, there it can communicate to the remote machine and in the remote
machine this is now again moved up the stack and at each layer that particular program or
whatever that module in that layer, he will pick out the message which has been sent by its peer
in the source. This is how communication will go on for all these protocols.

(Refer Slide Time: 40:25)

Sending a message is received on the receiving side, and then passed up through each layer and
each layer reads the corresponding header that means the corresponding header which has been
sent by its own peer on the source machine.
(Refer Slide Time: 44:11)

For example, we have two machines, and one machine is a DOS machine. That means some
DOS is an old operating system which used to run on the PCs and this was Microsoft DOS, PC
DOS, IBM DOS, etc. And Macintosh was a different company as I mentioned. So, these two can
still communicate, that means two different machines but this original data which was originated
from the application layer.

There is the presentation layer and this presentation layer will add its own header to the original
data. This whole thing becomes data for the next layer, namely, the session layer. Session layer
will add its own header so as I said, that the original data becomes fatter and fatter and fatter till
it comes to the physical layer. At the physical layer, data is actually sent in one hop, to the
destination machine. Here, as this whole fat packet travels up, each layer will strip off its
corresponding header which is coming from its peer in the source. That is how the each of the
peers can communicate and each of them has a virtual connection with the destination machine,
although the actual connection is the physical connection.
(Refer Slide Time: 44:34)

These are the seven protocol layers and now we are going bottom up. First is the physical link,
which is how to transmit the bits. Then we have the data link layer that is how to transmit
frames. Now what is the difference between transmitting bits and transmitting frames? We will
see to that; but a data link basically is a direct connection that is, how two computers which are
directly connected or two nodes which are directly connected to each other. How they will
communicate is the matter of data link. Computers need not be directly connected to each other
in order to communicate; they have to be connected to the network. If there are two machines A
and B or two nodes A and B which wants to communicate and there is no direct link between A
and B, so it may go as C D E and then B. Now, how do you know you have to go through C D
and E and then to reach B? Somebody has to keep track of the route. How to route the packets
over the entire network? That is the job of the network layer.
Then there is the transport layer that is how to send packets to the application. Packet is some
data which has been segregated and put into a packaged together. But the original application
layer need not have any concept of packet; somebody has to packetize all these data and there are
may be other functions in the transport layer. That is they make packets out of the data given by
the upper layer, this is a transport layer. The session layer, which manages the virtual connection
between in the application layer. It manages the virtual connection.

The presentation layer that means how we encode and decode messages that means two different
machines may have two different ways of encoding things. So, that is the job of the presentation
layer as well as the security comes here. Application, so whatever the actual job the human user
is interested in comes in the application layer.
(Refer Slide Time: 46:08)

The application layer contains a variety of protocols; it depends on what you are trying to do.
The various users may want to do various things; examples are; FTP, TELNET, SMTP, HTTP
etc. so these are names of some protocolsm By the FTP protocol, what u can do is you can
download files from another machine, by HTTP, you can look at web pages, surf etc. and SMTP
is used for sending mails. You may not have come across these protocols directly but the point is
whenever you are sending a mail, let us say from a UNIX machine or something, you are using
the SMTP protocol, so SMTP protocol has been built into that.
Similarly, even if you is TELNETing that means the TELNET sever, TELNET client etc, they
will be there in the machine as part of the OS. So, the application layer usually, request reliable
and cheap connection to its peer. That is what the application layer is concerned about: that it
must be connected reliably and cheaply to its peer. Some examples of peers we have given some
nodes given service etc.

(Refer Slide Time: 48:43)

Application layer hands this, whatever it has to send to the presentation layer which handles the
format of the data: Protocol conversion, data translation with its ASCII or may be some other big
or something and data compression, data encryption, these are handled in the presentation layer.
(Refer Slide Time: 50:00)

In the session layer, it allows the application on different computers to share a connection. We
will see about this connection later on. I go to the next layer we will see that I will packetize this
data but then so far as the user is concerned, he wants a continuous connection.

This connection is handled in the session layer so it can provide check points; that is if there is
come disruption, you can come back to it and get the original state back and then you can
retransmit if there is some distance dialogue control, who can transmit etc.
(Refer Slide Time: 50:22)

The transport layer: the basic function of the transport layer is to accept data from the layer
above, split it into smaller units if necessary. Now, why we need to split? That we will come to
later on. So these are the packets. Pass these to the network layer and ensure that the pieces all
arrive correctly and in the right order at the other end. For example, you have chopped them into
pieces; they may have become out of order. That is the business of the transport layer. That
should also be done in a cheap and efficient manner and etc. and the same thing applies.

(Refer Slide Time: 51:01)

There may be different types of transport services; there may be multiple protocols here. One
could be error-free, point-to-point channels or it could be a very cheap kind of channel or
broadcasting of messages to multiple destinations. So these are the different types of transport
service, which are possible.
(Refer Slide Time: 51:33)

Network layer decides on what route to take locally, so that the intended message ultimately
reaches the destination. It controls the broadcasting by essentially segregating the different
networks, etc. it handles technological mismatches so we will get into the details of this later on.
It does congestion control, it handles billing information etc.

(Refer Slide Time: 51:56)

(Refer Slide Time: 52:12)

The data link layer makes the physical link layer appear like a channel that is free of
transmission errors. Actually in the physical layer there may be error but the data link layer
handles this error correction etc. at the very lowest level. Finally, at the lowest level we have the
physical layer, where the data is actually transmitted as raw bits, etc.

(Refer Slide Time: 52:20)

(Refer Slide Time: 52:3)

The other layer is TCP by IP reference model, where the session and presentation layer are not
there. We have the application layer, transport layer, network layer or maybe something like the
data link or the physical layer.

(Refer Slide Time: 52:22)

So the rough correspondence is something like this; we have the TCP by IP model where we
have just this few and we have the OSI reference model with so many. So more or less, they
match the essential functions match. There are other kinds of protocols, which are also used but
the TCP by IP and OSI, are the most common. In the next class, we will be discussing the
different structures of networks. So, today we just had an abstract view of networks and in the
next class we will handle different structures.
(Refer Slide Time: 52:52)

Preview
Computer Networks
Prof: Sujoy ghosh
Indian Institute of Technology, Kharagpur
Lecture - 2
Network topology
Good day. In this lecture we will discuss the network topology. Now what do you mean by a
network topology?
(Refer Slide Time: 54:00)

(Refer Slide Time: 54:09)

We just have a quick recap of what we had learnt last time, basically a computer network, could
be a number of nodes which are connected by some communication links. There is some kind of
graph, this graph has certain structure so when we talk about the structure, this structure has an
implication about how will go about communicating as I said it is in general not feasible to have
one to one communication between each pair of nodes. That is not possible at all, so what kind of
structure? That is why the nodes connected in each manner is the subject matter of our discussion
today. A network may be represented as a graph, nodes representing computers or network
devices like switches, routers etc. and the links represent communication links. Modes of
communication may be broadcast or point to point. We have discussed this.
(Refer Slide Time 55:25)

First let us talk about the LAN topologies. LAN, the local area network and the local area
network topologies are three. They are very common; star, ring and a bus.

(Refer Slide Time: 55:49)

First of all, we first take up this bus topology and why do you take up a bus topology? This a
very simple kind of network which is based on a shared broadcast links. We still want to look at
point to point communication, that means there may be so many nodes a, b, c, d, etc., which are
connected to the network and a wants to communicate to b. so c, d, e are connected but a
specifically wants to communicate to b. Each pair of communicating nodes used a link for the
short time. So, a uses the link for the short time to send his message to b then may be c might
sends something to b or something like that. Other nodes ignore the communication. Since, this
is shared broadcast link all the nodes get the communication, that is not private in that sense. All
the nodes get that communication but they usually ignore this communication whereas b will
copy this for its own purpose. Now, they has to be a distributed protocol to decide who gets to
use the link. There has to be some protocol otherwise if a wants to communicate with b and b
wants to communicate c, as c wants to communicate with d their communications will go and
collide in that shared broadcast medium, so the communication of both the pairs of node will get
garbled. So by that I mean and sometimes actually that would happen that things will get garbled
but that is not what we.